Author: Aaron Gordon
Date: 19:20:44 12/30/03
Go up one level in this thread
On December 29, 2003 at 10:41:57, Slater Wold wrote: >On December 28, 2003 at 03:22:17, enrico carrisco wrote: > >>On December 28, 2003 at 02:40:15, Slater Wold wrote: >> >>>On December 28, 2003 at 01:36:50, enrico carrisco wrote: >>> >>>>On December 27, 2003 at 12:22:30, Slater Wold wrote: >>>> >>>>>On December 27, 2003 at 07:49:34, Aloisio Ponti Lopes wrote: >>>>> >>>>>>. >>>>> >>>>>Athlon 64 FX51 (2.2Ghz) = 1.64M nps >>>>>Opteron 2.2Ghz = 1.64M nps >>>>>Athlon 64 3200+ (2.0Ghz) = 1.50M nps >>>>>P4 3.6Ghz = 1.48M nps >>>>>AMD Athlon XP 3200+ (2.2Ghz) = 1.45M nps >>>>>P4 3.2Ghz Extreme Edition = 1.37M nps >>>>>Opteron 1.8Ghz = 1.35M nps >>>>>P4 3.0Ghz = 1.23M nps >>>> >>>>Did you run all these tests or have they been calculated another way? >>>> >>>>-elc. >>> >>>I ran the FX51, the P4 3.6, and the P4 3.0Ghz tests. >>> >>>The others were calculated based on SPEC. >> >>Reason I ask is because 1.45M nps for an XP 2.2GHz with an unoptimized (or not >>fully optimized) binary is a bit overstated in my opinion. 1.54M nps is what >>Aaron's 2.5GHz XP with an *optmized* binary gets. Since I have never seen >>anyone with a faster AMD binary than Aaron, I have to guess that the one you ran >>wouldn't be either. >> >>Thus, with Aaron's binary, an XP 2.2 should get ~ 1.36M nps. >> >>-elc. > >Aaron's 2.5Ghz was like 3% slower than a true Barton 2.2Ghz, on the perft bench >for Albert's distributed project. > >These tests were run with an optimized exe; Dann's. Mine was faster when using the same hash size. It (the Barton) was only faster with a much larger hash. Also, from what I've seen with chess engines the perft results had absolutely no correlation to actual chess playing speed, with my Athlon XP being faster than the Barton by a linear amount.
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.